The present invention relates to internal unbalance detection in capacitors using circuitry internal to the capacitor, and more particularly to the pre-detection of capacitor section failures within a capacitor. Even more particularly, the present invention relates to the detection of capacitor section failures within a capacitor using internal circuitry and, for example, a fiber optic output, wherein capacitor failure is detected upon, for example, a detected failure of a prescribed number of capacitor sections and capacitor failure is pre-detected, i.e., incipient capacitor failure is detected upon, for example, a detected failure of a predetermined lesser number of capacitor sections. Even more particularly the present invention also relates to detection of incipient capacitor failure with a widely varying voltage on the capacitor, without the need to recalibrate the detection system to account for the widely varying voltage.
Present day operational equipment and research devices require the use of large capacitors connected in banks, and in some instances employ capacitor sections connected in commonly-housed capacitors. As used herein the term capacitor section refers to a single capacitor element, e.g., in general, a pair of conductive structures, e.g., plates, separated by a dielectric material. The term capacitor refers to a functional electrical component made up of at least one capacitor section, and having at least two externally accessible electrodes through electrical connection can be made to the one or more capacitor sections. The term capacitor module refers to that portion of, for example, a high voltage device, such as an accelerator or a laser, made up of at least one capacitor, and associated hardware. By high voltage it is meant a device operating at, for example, more than 5 kilovolts, e.g., 10 kilovolts.
Standard practice for high voltage capacitors dictates that each capacitor consists of several parallel series of high voltage capacitor sections in a sealed casing (or housing).
A short circuit (which is a typical failure mode for high voltage capacitor sections in high voltage capacitors) in one or more of the high voltage capacitor sections may result in a rapid increase in heat in the capacitor and may result in an explosion and ensuing fire. Since the capacitors are made of series-connected capacitor sections and because each capacitor section in the series may be operating below its maximum voltage rating, e.g., at 80% of its maximum voltage rating, damage as a result of a short circuit in one high voltage capacitor section may be temporarily held off by other capacitor sections in the series (because they will operate closer to their maximum voltage rating, but not exceed their maximum voltage rating).
As a result, it is typically difficult to even determine if a short circuit in an individual high voltage capacitor section within the sealed housing of the capacitor has occurred, before a catastrophic failure occurs. The same situation occurs for capacitors made up of capacitor sections that fail in an open circuit mode utilizing self-clearing electrodes or internal fuses. Once a single high voltage capacitor section within the capacitor fails, however, a subsequent failure of another high voltage capacitor section within the capacitor may result in remaining capacitor sections operating at or above their maximum voltage ratings. At this point one can expect the remaining high voltage capacitor sections to fail rather quickly, resulting in failure of the capacitor. Thus, while it is important that a short circuit in an individual high voltage capacitor section in a capacitor be detected and corrected as soon as possible, before explosion or fire, to prevent damage to equipment or injury to personnel, no satisfactory means of such detection is commercially available.
One prior art approach to detecting faulty capacitor sections within capacitors is to monitor liquid dielectric pressure within the capacitors. Because the liquid dielectric pressure changes abruptly, due, for example, to gas generation or when the temperature inside the capacitor rises abruptly due, for example, to a short circuit, measuring the liquid dielectric pressure provides an indication that a short circuit has occurred. Specifically, when a high voltage capacitor section fails, the high voltage capacitor section typically has a short arcing between its plates causing a build up of heat and/or gas, and in turn, a build up of pressure in the liquid dielectric, which can be measured at a pressure valve. Typically an interlock mechanism is then used to remove the capacitor from service, thus shutting down whatever system is being used with the capacitor.
Advantageously, this type of detection does not require any external electrical connections, which can pose a significant problem when capacitors are themselves series connected or operating at high voltages. (This problem arises because while any electrical outputs from short-detecting circuitry may only be a few volts in the capacitor, the voltage above "earth ground" can be on the order of several kilovolts or more.)
Unfortunately, liquid dielectric pressure detection requires long periods for detection due to a significant time required to build up sufficient heat and pressure within the liquid dielectric to result in a detection since the system must be able to handle normally expected changes in temperature due to changes in the ambient operating temperature or the internal temperature rise due to operating conditions. Thus, the time required to detect a capacitor failure by the so-called pressure method often results in explosion because of the destruction of other elements due to over-stresses leading to a short circuit in the over-stressed sections. And, even when a capacitor failure is detected, such detection is generally considered a detection of imminent failure, and thus an interlock mechanism or the like must be used to take the capacitor offline, thus shutting down whatever equipment is being used with the capacitor. Also, the pressure interrupters sometime yield a false signal due to changes in pressure inside the capacitor associated with operating of conditions other than failure. These "unscheduled" shutdowns can be quite costly in, for example, experimental accelerators, as valuable experiments can be spoiled and valuable accelerator time lost.
An alternative approach involves monitoring voltage across each individual capacitor section within the capacitor. An example of this approach is highlighted in U.S. Pat. No. 3,125,720 (Swift) and in U.S. Pat. No. 4,805,063 (Kataoka, et al.). Each of these patents describes one or more approaches for detecting a failure by monitoring voltage across a voltage divider set in parallel with the series connected capacitor sections within the capacitor. Both of these patents, however, describe approaches that are unsuitable in extremely high voltage environments due to the requirement of electrical connections between the individual capacitor sections inside the housing of the capacitor to external unbalance detection circuits outside the housing of the capacitor, through the housing of the capacitor. As mentioned above, these electrical connections, in practice, can be of significant voltage potentials above earth ground, posing an extreme risk of arcing and risking damage to sensitive monitoring and experimentation hardware.
As pointed out above, another problem encountered in applications with which high voltage capacitors are used is that such applications require that the capacitors remain online continuously and not be unexpectedly switched out of service, such as, for example, in the event of an interlock mechanism switching the capacitor out of service upon detection of a capacitor failure. Examples of these types of systems include experimental accelerators used for the refinement of nuclear fuel, such as at Los Alamos National Laboratories, wherein a shutdown as a result of capacitor failure can result in spoiling very expensive experiments, and loss of precious accelerator time. Accordingly, prior art systems such as used in power line applications, that automatically switch capacitors out of operation using an interlock mechanism when over-voltages are detected, or that remove capacitors from operation, such as through series fuses when short-circuits are detected, are expensive and often unacceptable in many real-world applications in which high voltage capacitors are employed.
Thus, what is needed is an unbalance detection approach in which early detection (or pre-detection) of incipient capacitor failure can be made, without shutdown of the capacitor, unless a catastrophic failure, such as an explosion, of the capacitor is imminent. In this latter case, it would remain important to switch the capacitor off using, for example, an interlock mechanism, such as is known in the prior art, and for that matter switch remaining power circuits off, so as to prevent catastrophic failure, fires, and the like. However, it is also very desirable and preferable to allow capacitors with early-detected minor failures to continue to operate until repair can be scheduled and effected without interrupting valuable operations and experimentation.
The present invention advantageously addresses the above and other needs.